The Significance of Paced Electrogram Fractionation in Hypertrophic Cardiomyopathy
A Prospective Study
Background Increased duration of paced right ventricular (RV) electrograms in hypertrophic cardiomyopathy has been shown in 37 patients to correlate with the risk of ventricular fibrillation (VF). The changes in electrogram duration with pacing stimulus prematurity discriminated patients into three groups: VF survivors, an intermediate group with either nonsustained ventricular tachycardia (NSVT) on ambulatory monitoring or a family history of sudden death (FHSD), and those with none of these risk factors (noRF) for sudden death (SD). The consistency of these original groups has been tested prospectively in a further 64 patients.
Methods and Results Of 64 patients with hypertrophic cardiomyopathy, 3 had documented VF, 1 had witnessed SD and is assumed to have had VF, 25 had NSVT, 21 had FHSD, and 14 had noRF. Nineteen patients had syncope. They were studied by pacing one RV site with a decremental sequence and recording high-pass filtered electrograms from three other RV sites. The delay of each fractionated potential in the electrogram was determined relative to a pacing stimulus of increasing prematurity. These measurements were repeated by pacing each ventricular site in turn. The electrograms were characterized by two parameters: the extrastimulus coupling interval (S1S2) at which delay increased by more than 0.75 ms/20 ms decrease in S1S2 interval and the change in electrogram duration between an S1S2 of 350 ms and ventricular effective refractory period. The 4 VF patients had a mean increase in electrogram duration of 16.1 ms and an increase in delay at a mean S1S2 of 368 ms. Three VF patients were within the original VF group, while only 6 of 60 non-VF patients were within this group, discriminating between VF patients and the remainder (P<.007). The 14 noRF patients had a mean change in electrogram duration of 4.5 ms and an increase in delay at a mean S1S2 of 301 ms. Eleven patients were within the original noRF group, and only 8 of the remaining 50 patients also were within the noRF group, discriminating between the noRF patients and the remainder (P<.0005). Most of the NSVT and FHSD patients were between the original VF and noRF groups, with 5 of 25 NSVT and 1 of 31 FHSD patients in the original VF group. There was no relation between syncope and electrophysiological characteristics. Programmed electrical stimulation (PES) was performed in the first 15 patients of this study. Of the total 52 patients from the original and current studies, PES identified 2 out of 6 VF patients, and there was no correlation between VF inducibility and intraventricular conduction delay.
Conclusions These data are consistent with the original VF and noRF groups. Most patients with FHSD or NSVT were between these groups. Pooled data from the original and current groups (n=101) allow definition of a new VF group, which includes all patients with VF (n=9), 8 of 30 patients with VT, and 3 of 31 patients with FHSD. This new group may be used as a criterion for implantable cardioverter-defibrillator implantation in a prospective trial of the technique for the prediction of SD.
Hypertrophic cardiomyopathy (HCM) is an important cause of sudden death (SD) in adolescents and young adults.1 2 The annual SD rate in HCM reaches a maximum of between 4% and 6% per year in adolescence and early adulthood,3 4 falling to between 4% and 1% per year in later life.5 6 Since the original suggestion that SD in HCM was due to arrhythmias,7 there have been some confirmatory ambulatory ECG recordings of ventricular arrhythmias causing SD.8 9 Also, there is considerable circumstantial evidence that nonsustained ventricular tachycardia (NSVT), recorded using ambulatory ECGs, predisposes toward SD,10 11 implying an arrhythmic mechanism. However, the electrophysiological mechanisms of SD in HCM are poorly understood, and this has hampered investigation of patients with the disease. Therefore, a major goal in the management of HCM is to demonstrate the presence or absence of an arrhythmic substrate in a particular patient. This would permit prophylactic implantable cardioverter-defibrillator (ICD) implantation in high-risk patients and possibly avoid implantation in an unacceptably high number of individuals who will not subsequently develop life-threatening arrhythmias.
One predictor of SD, NSVT detected on ambulatory monitoring, carries a 23% risk of SD within 3 years.10 11 However, this has too low a positive predictive accuracy to justify ICD implantation in the whole group of patients with this marker. A similar argument applies to patients with a family history of sudden death (FHSD) or with syncope,3 12 13 which, while they are risk factors for SD, have low positive predictive accuracies. In this study, patients with syncope were not treated as an independent group but were treated as having ventricular fibrillation (VF), NSVT, FHSD, or none of these risk factors (noRF). However the interrelations between syncope and the electrophysiological characteristics of these patients are considered in the “Discussion.”
We reported in an earlier study a new technique to analyze the inhomogeneity of intraventricular conduction in the right ventricular myocardium of patients with HCM.9 The technique was designed to detect the putative effects of myocardial disarray and is based on the hypothesis that there are a number of discrete conduction paths throughout the myocardium. These paths were thought to arise due to several features of disarray14 : variations in fiber diameter leading to altered local conduction velocity and refractoriness and anisotropic conduction due to increased fibrosis. These changes were predicted to cause dispersion of conduction velocity through the ventricular myocardium and therefore create one component of the substrate for reentrant arrhythmias.
This dispersion of conduction velocity was demonstrated clinically by pacing at one site in the right ventricle and recording electrograms at three other sites. The basis of the technique is to analyze the change in delays of individual recorded electrogram potentials (fractionation), which presumably reflect depolarization of fibers close to the recording electrodes, in response to extrastimuli of increasing prematurity. Using this analysis, it was found that 4 patients who had been resuscitated from VF and 1 other patient who subsequently suffered a documented VF arrest had maximal electrogram lengthening, and this occurred at longer extrastimulus coupling intervals than in other patients. By contrast, control subjects and noRF patients showed the least change in electrogram morphology with premature stimulation. Patients with either NSVT on ambulatory ECG monitoring or FHSD had results that spanned the range from VF survivors to control subjects. Therefore, three electrophysiological groups could be established: the high-risk VF group with maximum fractionation, the noRF group with least fractionation, and an intermediate group composed of patients with either NSVT or FHSD.
The purpose of this study was to compare the results obtained from a further 64 patients with HCM with the predictions of risk obtained on the basis of the original 37 patients and to discuss the implications of these results for the management of high-risk patients, in particular those with NSVT on ambulatory monitoring.
Sixty-four consecutive patients (the current group) have been studied, and their results are presented as new data, which will be compared with the published results from 37 patients studied in the original study (the original group). Thirty-eight were male. Six patients were aged less than 20 years, 15 were between 20 and 29 years, 17 were between 30 and 39 years, 18 were between 40 and 49 years, and 8 were older than 50 years. HCM was diagnosed on the basis of characteristic clinical, electrocardiographic, and echocardiographic features. All except 3 patients had maximum left ventricular (LV) wall thicknesses in excess of 15 mm in the absence of systemic disease that could lead to ventricular hypertrophy. The 3 remaining patients had maximum LV wall thicknesses of 13 to 15 mm, with an abnormal ECG and HCM present in first-degree relatives. Four patients had left ventricular outflow tract gradients measured by echo Doppler of between 30 and 60 mm Hg, and 5 had gradients greater than 60 mm Hg. Three patients had limitation of New York Heart Association (NYHA) grade 3, 36 patients had grade 2, and the remainder had grade 1. Twelve had exercise-induced chest pain.
Three patients had been resuscitated from electrocardiographically proven VF. One further patient, who had FHSD, was witnessed to collapse and die suddenly 4 weeks after he had been studied. As no other cause of death was found at postmortem examination, he is included in the VF group. Twenty-five patients had NSVT on ambulatory monitoring, which was defined as more than 3 beats of a broad-complex tachycardia at a rate of more than 120 beats per minute recorded during 48 hours of ambulatory ECG monitoring. Twenty-one patients had a history of SD in first-degree relatives, and 14 had neither of these risk factors (noRF). Nineteen patients had syncope. Of these, 3 had VF, 7 had NSVT, 4 had FHSD, and 5 had noRF.
The patients gave written informed consent for the study, the protocol having been approved by the Wandsworth District Ethics Committee. They were premedicated with oral lorazepam (2 to 4 mg) 1 hour before the study and were sedated during the study with intravenous diamorphine and diazepam. Prochlorperazine 12.5 mg was given intravenously at the start of the study. Bipolar electrode catheters were positioned via the femoral veins in four sites within the right ventricle at the apex, midseptum, inferior wall, and outflow tract. One site in the right ventricle was paced with a constant rate sequence, with an extrastimulus inserted every third beat. The high right atrium was paced simultaneously with the ventricle to avoid fusion between paced and sinus beats. The extrastimulus coupling interval (S1S2 interval) was reduced in 1-ms steps from 450 ms down to the ventricular effective refractory period (VERP). During the sequence, bipolar electrograms were recorded digitally from the three other right ventricular (RV) sites at 1 kHz with 12-bit accuracy after conditioning with a single-pole, high-pass filter (−3 dB=50 Hz) and a 4th-order Bessel antialiasing filter. The sequence was delivered from each electrode in turn, and recordings were made from the remaining catheters. Thus, there were four runs, each consisting of 200 to 250 electrograms in response to extrastimuli, measured from three ventricular sites, yielding a total of 12 sets of electrograms.
After these pacing runs, programmed electrical stimulation (PES) was performed from the RV apex up to stage 8 of the Wellens protocol in the first 15 patients of this series. The reasons for discontinuing PES are described later in the “Discussion.”
The basis of the technique is to analyze how these electrograms change in response to an extrastimulus, and this is illustrated in Fig 1⇓. The potentials in each electrogram, which presumably correspond to the activation of fibers close to the recording electrode, were emphasized using a single-pole, digital, zero-phase, high-pass filter (−3 dB=150 Hz). The amplitude distribution of the signal was then determined between 150 and 200 ms after the pacing spike (Fig 1a⇓). All fractionated potentials between the pacing spike and 150 ms with an amplitude greater than twice the range of the noise were detected (Fig 1b⇓). The delay of each potential in a particular electrogram was plotted against the S1S2 interval at which they were recorded, and, when this is done for each electrogram, an intraventricular conduction curve was obtained (Fig 1c⇓) that described the conduction properties of the myocardium between the stimulating and recording electrodes. Each curve, generated in response to extrastimuli, was characterized by two parameters: (1) the S1S2 interval at which electrogram delay started to increase and (2) the change in the electrogram duration between an S1S2 interval of 350 ms and VERP. These parameters were determined by fitting interpolating functions, cubic splines,15 to the delay of the earliest and latest components of the high-pass–filtered electrogram as a function of S1S2 interval (Fig 1d⇓). The change in electrogram duration was measured directly from the two interpolating splines, and the S1S2 interval at which the delay started to increase by more than 0.75/20-ms reduction in S1S2 interval was measured by differentiating the spline fitted to the latest electrogram component. Finally, the mean of these measurements from each curve was obtained (Fig 1e⇓) and used as an independent observation to describe the intraventricular conduction in a particular patient. These plots are shown in Figs 2 through 5⇓⇓⇓⇓. In each figure, line A is the discriminant line that separated the VF group from the remainder, calculated on the basis of the original 37 patients, and line B is the line that separated the noRF group and control subjects from the patients with risk factors, again calculated from the same 37 patients.
The results of this study are plotted on scatterplots (Figs 2 through 6⇑⇑⇑⇑⇓) of mean change in electrogram duration between 350 ms and VERP against the S1S2 coupling interval at which delay started to increase. Those patients who suffered a VF arrest (n=4) showed, as a group, the greatest fractionation, with a mean S1S2 at which delay increases of 368 ms and a mean increase in electrogram duration of 16.1 ms. By contrast, the noRF group (n=14) had minimal fractionation (mean S1S2 of increased delay, 301 ms; mean change in electrogram width, 4.5 ms). These data are plotted in Fig 2⇑, where there is clear separation between the noRF group and the VF group. Fig 3⇑ shows a similar scatterplot of 25 patients with NSVT on ambulatory monitoring. There is a spread across the entire range of patients, with 5 patients to the right of line A (that is, in the high-risk group) and 3 patients to the left of line B (with the noRF group), while the majority (n=17) are between these two groups. Fig 4⇑ is a scatterplot of the remaining 21 patients with a malignant family history. Again, this group spans the range from the VF group to the noRF group, with 1 patient in the VF group and 5 in the noRF group. Line A discriminates the current VF group from the remainder (P<.007, Fisher’s exact test), and line B discriminates the current noRF group from the remaining population (P<.0005, Fisher’s exact test).
This work was based on the hypothesis that disarray in HCM would produce a potential substrate for ventricular arrhythmias, with several slowly conducting paths within the ventricles, and that the risk of arrhythmias would be correlated with the magnitude of the intraventricular conduction disturbance. This hypothesis has previously been tested and reported in 37 patients,9 where it was found that VF survivors were characterized by increased electrogram duration with extrastimulus prematurity and increasing delay at long S1S2 coupling intervals when compared with the remaining population. In the present study, using the criteria defined in the initial study, there is a significant discrimination between VF survivors and the remaining population. There is also significant discrimination between noRF patients and the remaining population with either VT, FHSD, or VF. Therefore, the original hypothesis that there is an intraventricular conduction disturbance in high-risk patients has been strengthened by testing it against a new population of patients.
An important feature of this study is that there were 25 patients with NSVT, in contrast to the small sample reported (n=5) in the original study. Patients with NSVT during ambulatory monitoring had an increased SD rate of approximately 23% within 3 years.10 Despite this high annual death rate, the clinical interpretation of NSVT is problematic because of its low predictive accuracy for SD. This study shows that there is a wide range of variation of intraventricular conduction disturbance in NSVT, ranging from the VF to the noRF groups, suggesting that there is a spectrum of risk of SD in NSVT, which is consistent with clinical experience. There was a similar spread of patients with FHSD who showed a wide range of results ranging from the VF to noRF groups. Again, this result is consistent with clinical experience that not all patients from families with SD die suddenly.
Nineteen patients had syncope, and their positions are shown in Fig 5⇑. Their results ranged from the VF group to the noRF group, and there was no significant clustering in one particular risk category. Using pooled data of 101 patients from the original and current study groups, there was a significant bias toward syncope in 6 of 9 VF patients compared with 30 of 92 patients without VF (P<.05, Fisher’s exact test). Again, the 30 patients with syncope were equally divided between the NSVT, FHSD, and noRF groups. The presence of syncope in patients with a wide range of electrophysiological characteristics (particularly VF patients) is consistent with the idea that it is a trigger for arrhythmic SD in a susceptible individual. This concept is important because it potentially allows distinction to be made in patients with syncope between those with a major arrhythmic substrate, who are presumably at high risk of SD, and those with a minor substrate, who may be at less risk of SD.
PES was performed in a total of 52 consecutive patients, using 15 patients in this series and 37 patients from the previous series. Twelve patients developed VF or polymorphic VT requiring defibrillation (a positive result), and 40 patients did not. There is no obvious relation between the induction of VF and the degree of fractionation, as shown in Fig 6⇑. Two VF patients were inducible, while 4 were not, including 1 patient who was studied before a documented VF arrest. One of the 2 patients who were inducible had monomorphic VT with a cycle length of 360 ms recorded before developing VF on a treadmill; however, only polymorphic VT was induced rather than the clinical tachycardia. The 10 non-VF group patients who were positive with PES have remained alive and well without arrhythmic events for a follow-up of 2 to 4 years. These data suggest that PES measures a different electrophysiological characteristic to intraventricular conduction and raise the important question as to whether induced arrhythmias in patients with HCM bear any relation to their potentially fatal clinical arrhythmias. The relative merits of PES and intraventricular conduction can only be tested by prospective trials of the two techniques; however, using our data, the positive predictive value of PES is .166 and its negative predictive value is .9. This suggests that the ability of PES to discriminate patients at risk of VF is low, which is the conclusion reached by Kuck et al,16 based on results similar to ours.
Would a more aggressive PES protocol have given a different result? The incidence of polymorphic VT rises steeply when a third extrastimulus is added,17 and hence the likelihood of a positive-result PES is increased. In our series, it appears unlikely that more aggressive stimulation would have selectively produced a positive result in the VF survivors and not in non-VF patients. A stimulation protocol with up to 3 extrastimuli from the right and left ventricles induces VF in 70% of VF survivors.18 Out of 230 patients with HCM in this series, 82 had a positive result with PES, and 14 of these patients, together with 3 who had a negative result with PES, subsequently had a cardiac event. However, of the remaining 213 patients who did not have a cardiac event, 68 (32%) had a positive PES result. It is the false-positive rate of PES that raises doubts about its suitability for screening patients with HCM, since it suggests that any intervention, such as an ICD, would be necessary in a large group of patients who do not require them.
Finally, we regard any procedure in patients with HCM with an end point of VF as potentially hazardous. The last patient in our PES series developed prolonged VF that converted to asystole with multiple defibrillation discharges, which was refractory to pacing from any ventricular site, and he was resuscitated with great difficulty. This case made us critically examine our PES data, and because of its inability to distinguish VF patients from the remaining population, we discarded PES as a means of evaluating the risk of SD in patients with HCM.
The intraventricular conduction pacing protocol induced VF in 12 patients, including 4 VF patients, which is half the induction rate of PES. However, this rate is undesirable, since the object of measurement of intraventricular conduction is to evaluate a patient without inducing an arrhythmia. There is no consistent marker for the development of VF during a pacing run. In 2 patients, there was an increase of more than 20 ms between the intrinsic deflections in different channels with premature stimulation before the induction of VF. In 3 patients, VF occurred shortly after the point at which delay started to increase. In the remainder there was no obvious electrophysiological event, and in 6 patients this occurred close to VERP without any major degree of fractionation.
This raises the important question of how to modify the protocol to minimize the risk of VF. This study used the same protocol as the initial study for obvious reasons of comparability. However, the basic technique can clearly be modified. One modification is to increase the step by which the S1S2 interval is shortened. Curves using 2-ms steps are now under investigation, which will halve the study time and may reduce the risk of arrhythmias. Again, the curves can be generated using single extrastimuli inserted during sinus rhythm or during constant rate atrial pacing. Finally, it may be possible to detect the point at which delay starts to increase during the run and terminate the run at that point while gaining information about the change in electrogram duration using a short run near the VERP. Also, the arrangement of electrode catheters may not be ideal. More information may be available by recording from a multipolar catheter placed on the RV septum and by pacing the LV apex. One-centimeter–spaced catheters were used in this study because the technique was designed to measure local inhomogeneity of conduction, which requires sampling a significant volume of myocardium. However, to answer other questions, such as the change in conduction velocity of the intrinsic deflection, a smaller interelectrode spacing would be appropriate. Therefore, while this study has confirmed the results of the smaller initial study, the basic technique requires further development to reduce the incidence of induced arrhythmias and to reduce its complexity.
The pooled data from the original and current groups of 101 patients suggest one approach to establishing criteria for ICD implantation on the basis of intraventricular conduction measurements. Fig 7⇓ shows the positions of the 9 patients with VF, 30 patients with NSVT taken from the pooled group, and 7 control subjects (3 men and 4 women; ages 17 to 67 years). (All had normal resting ECGs and echocardiograms, 2 had AVNRT, 2 had palpitations with a normal electrophysiological study, 1 had paroxysmal atrial fibrillation, and 2 had syncope with a normal electrophysiological study.) Line C is constructed to discriminate the VF population from the remainder while including all patients with VF.19 Eight of 30 patients with NSVT are included in this group, and this proportion is comparable to the number of patients with NSVT who die suddenly within 3 years.10 Also, 3 of 31 patients with FHSD are in this group, which is again comparable with the 3-year mortality of this marker. Despite the likelihood that there are many initiating mechanisms of SD in HCM, including ischemia and syncope,3 13 20 the use of this line as the criterion for ICD implantation, in the absence of any identifiable or treatable trigger, may give a rational approach to prospective testing of the efficacy of intraventricular conduction measurements as a marker of SD in patients with NSVT.
The results of this study confirm that there is an electrophysiological abnormality in patients with VF and HCM and suggest a method of assessment of risk in three problematic groups: those with NSVT, FHSD, and syncope. The validity of this approach can only be tested by a prospective trial of the technique of intraventricular conduction measurement to predict the use of ICDs. However, the results obtained from 101 patients in total suggest that patients who have similar characteristics to VF survivors should not be left unprotected and that such a prospective trial is justified.
This study was supported by the British Heart Foundation (A.K.B.S., A.A.G.) and by Fédération Française de Cardiologie (N.S.).
- Received August 25, 1994.
- Revision received December 1, 1994.
- Accepted December 18, 1994.
- Copyright © 1995 by American Heart Association
Maron BJ, Roberts WC, McAllister HA, Rosing DR, Epstein SE. Sudden death in athletes. Circulation. 1980;62:218-229.
McKenna WJ, Deanfield JE. Hypertrophic cardiomyopathy: an important cause of sudden death. Arch Dis Child. 1984;59:971-975.
Maron BJ, Roberts WC, Epstein SE. Sudden death in hypertrophic cardiomyopathy: a profile of 78 patients. Circulation. 1982;65:1388-1394.
Kofflard MJ, Waldastein DJ, Vos J, ten Cate FJ. Prognosis in hypertrophic cardiomyopathy observed in a large clinic population. J Am Coll Cardiol. 1993;72:939-943.
Saumarez RC, Camm AJ, Panagos A, Gill JS, Stewart JT, de Belder MA, Simpson IA, McKenna WJ. Ventricular fibrillation in hypertrophic cardiomyopathy is associated with increased electrogram fractionation. Circulation. 1992;86:6467-6474.
McKenna WJ, England D, Doi YL, Deanfield JE, Oakley CM, Goodwin JF. Arrhythmia in hypertrophic cardiomyopathy, I: influence on prognosis. Br Heart J. 1981;46:168-172.
McKenna WJ, Harris L, Deanfield J. Syncope in hypertrophic cardiomyopathy. Br Heart J. 1982;47:177-179.
Maron BJ, Fananapazir L. Sudden cardiac death in hypertrophic cardiomyopathy. Circulation. 1992;85(suppl I):I-57-I-63.
Davies MJ. The current status of myocardial disarray in hypertrophic cardiomyopathy. Br Heart J. 1984;51:361-374.
Numerical Algorithms Group Scientific Subroutine Library. Mark 14. Routines E02BEF and E02BBF. Oxford, England: NAG Ltd; 1991.
Kuck KH, Kunze KP, Nienaber CA, Costard A. Programmed electrical stimulation in hypertrophic cardiomyopathy: results in patients with and without cardiac arrest or syncope. Eur Heart J. 1988;9:177-185.
Wellens H, Brugada P, Stevenson WG. Programmed electrical stimulation of the heart in patients with life-threatening arrhythmias: what is the significance of induced arrhythmias and what is the correct stimulation protocol? Circulation. 1985;72:1-7.
Fananapazir L, Chang AC, Epstein SE, McAreavey D. Prognostic determinants in hypertrophic cardiomyopathy: prospective evaluation of a therapeutic strategy based on clinical, Holter, hemodynamic, and electrophysiological findings. Circulation. 1992;86:730-740.
Hoel PG. Introduction to Mathematical Statistics. New York, NY: John Wiley & Sons; 1971:181-186.